University of Pennsylvania Working Papers in Linguistics Volume 20 Issue 1 Proceedings of the 37th Annual Penn Article 10 Linguistics Conference 2014 Looking Back and Looking Forward: Anaphora and Cataphora in Italian Emily Fedele University of Southern California Elsi Kaiser University of Southern California Follow this and additional works at: https://repository.upenn.edu/pwpl Recommended Citation Fedele, Emily and Kaiser, Elsi (2014) "Looking Back and Looking Forward: Anaphora and Cataphora in Italian," University of Pennsylvania Working Papers in Linguistics: Vol. 20 : Iss. 1 , Article 10. Available at: https://repository.upenn.edu/pwpl/vol20/iss1/10 This paper is posted at ScholarlyCommons. https://repository.upenn.edu/pwpl/vol20/iss1/10 For more information, please contact [email protected]. Looking Back and Looking Forward: Anaphora and Cataphora in Italian Abstract Pronoun interpretation is central for comprehension. Prior work focused mostly on anaphora, where pronouns refer to previously-mentioned antecedents. Less research is on cataphora, where antecedents follow pronouns. Existing work suggests cataphora triggers an active-search mechanism: The parser actively searchers for a syntactically-licenses antecedent. Our results on Italian null and overt subject pronouns show that both processing constraints (“impatient parser”) and the grammatical properties of referring expressions contribute to the outcome of reference resolution; parsers try to “discharge” unresolved pronouns when encountered first (i.e., cataphora) due to a processing load of keeping an unresolved pronoun in memory, even if this goes against grammar specific properties. Furthermore, in line with related research, we find that strong grammatical principles (Binding Theory) are powerful enough to “block” processing effects, contributing to our view of how different components of language processing interact. This working paper is available in University of Pennsylvania Working Papers in Linguistics: https://repository.upenn.edu/pwpl/vol20/iss1/10 Looking Back and Looking Forward: Anaphora and Cataphora in Italian Emily Fedele and Elsi Kaiser* 1 Introduction The question of what influences people’s interpretation and use of different referring expressions (e.g., she, him, it) has been investigated from a range of research perspectives. Prior research has largely focused on identifying what information hearers use to interpret pronouns. Pronouns are semantically under-informative forms and must receive their interpretation from the surrounding context; they do not provide enough information on their own to identify a referent, and yet they are frequently used in language without difficulty. Researchers have proposed various strategies that govern the search for a possible antecedent. This includes attention-driven approaches (e.g., Ariel 1990, see also Gundel, Hedberg and Zacharski 1993 for related work), which are focused on how accessible different entities are in the minds of the speaker and hearer. Simplifying things somewhat, it is commonly agreed that the more accessible the antecedent is, the more likely it is to be referred to using a pronoun. Researchers have also noted that pronouns tend to prefer anteced- ents in subject positions, perhaps due to subjects being more accessible (Crawley, Stevenson and Kleinman, 1990, among others), while others observed that the subject preference may in fact be part of a Parallel Structure preference (Smyth 1994, Chambers and Smyth 1998), where pronouns are resolved to antecedents that occupy a matching argument position. Recently, Kehler (2002), Rohde (2008), and colleagues have argued in favor of a different approach, which regards pronoun resolution as a side effect of the more general coherence-establishing processes that language us- ers engage in (see Hobbs 1979). In this paper, we report two experiments investigating how the human language processing mechanism comprehends different kinds of linguistic expressions, in particular null and overt pro- nouns in Italian. We investigated both anaphora, where a pronoun comes after its antecedent (1a) and cataphora where the pronoun linearly precedes its antecedent (1b) (examples from Kennison, Fernandez and Bowers 2009). (1) a. Anaphora: After Ted arrived, he asked for a cup of coffee b. Cataphora: After he arrived, Ted asked for a cup of coffee 2 Previous Work on Pronoun resolution 2.1 Existing Work on Cataphora Prior work on pronoun interpretation has focused on identifying strategies affecting the search for a pronoun antecedent mostly looking at anaphora. There is considerably less work on cataphora. In this section we briefly summarize some of the relevant studies on cataphora. In one of the earliest experiment studies to this topic, Cowart and Cairns (1987) observed a strong preference to interpret a cataphoric pronoun as referring to the first possible noun phrase that is encountered after the cataphor. Van Gompel and Liversedge (2003) conducted an eye- tracking reading study that builds on this work. Using sentences like (2), they manipulated wheth- er the gender of the cataphoric pronoun matches the gender of the subject noun (encountered first) or the object noun (encountered later). (2) When she was fed up, the (girl/boy) visited the (girl/boy) very often The eye-movement patterns show that comprehenders try to link the cataphoric pronoun to the first available noun (the subject), rather than to the second noun (the object). Broadly speaking, van Gompel and Liversedge’s results suggest that the parser actively searches for possible ante- cedents for the cataphoric pronoun, anticipating the upcoming main clause subject pronoun, and creates a referential dependency with that position. Thus, their results are in line with the earlier findings of Cowart and Cairns: Both point to an “impatient parser” that tries to resolve the pro- U. Penn Working Papers in Linguistics, Volume 20.1, 2014 82 EMILY FEDELE AND ELSI KAISER noun as early as possible. In related work, Kennison, Fernandez and Bowers (2009) found that anaphoric pronouns are processed faster than cataphoric ones, which they attribute to an expectation for coreference with an already mentioned antecedent: Anaphoric pronouns do not have to wait for an upcoming refer- ents, whereas cataphoric pronouns do. Other studies looking at cataphora have highlighted how the anaphor/cataphor distinction is much more complex when we consider languages other than English. Kazanina and Phillips (2010), investigating Russian, show that there is a time-course distinction between two different kinds of constraints affecting cataphoric pronouns: (i) a Russian specific constraint with cataphora, banning coreference between a main clause subject and subordinate clause pronoun when the con- nective is poka “while” and, (ii) Principle C of the Binding Theory, a grammatical constraint rul- ing out coreference between a pronoun and any referring expression that it c-commands (Chomsky 1981, Reinhart 1983). Principle C essentially fully blocks the consideration of any antecedent that would violate this constraint, while the Russian specific constraint acts as a “delayed” filter, where a restricted antecedent is originally considered before being ultimately filtered out of consideration. Summarizing, existing research on cataphora has identified specific differences in processing of anaphoric and cataphoric pronouns: Anaphoric pronouns are processed more quickly than cata- phoric ones due to a preference for coreference with already mentioned referents, whereas cata- phora involves an “active-search mechanism” that aims to identify the possible antecedent as soon as possible. In addition, Kazanina and Phillips (2010) found language-particular vs. more univer- sal constraints can differ in how effective they are in guiding reference resolution. 2.2 Reference Resolution in Italian: Referential Biases of Null and Overt Pronouns The impact of the anaphor/cataphor distinction in Italian is central to the ongoing debate concern- ing what information comprehenders use to guide pronoun interpretation. Italian has both null and overt pronouns: In an anaphoric configuration, null pronouns typically refer to preverbal subjects and overt pronouns typically refer to objects, as in (3). (Example from Carminati 2002). It has also been suggested that overt pronouns can potentially signal a topic shift. (3) Marioa ha telefonato a Giovannib, quando NULLa/luib aveva Mario has telephoned to Giovanni when null/he had Appena finite di mangiare just finished of to eat “Mario called John, when he just finished eating” Existing work on Italian subject pronouns has mostly focused on anaphoric configurations. In her influential dissertation, Carminati (2002) conducted a series of questionnaire studies and self- paced reading experiments with Italian speakers focusing on the referential biases on null and overt pronouns in subject position. Based on the apparent bias of null pronouns to prefer subject antecedents and overt pronouns to prefer object antecedents, Carminati argues in favor of a struc- tural approach to subject pronoun processing, the Position of Antecedent Hypothesis (PAH). Ac- cording to the PAH null pronouns refer to a structurally prominent antecedent in a [Spec IP] posi- tion, and overt pronouns refer to an antecedent lower in the clause structure. Recent evidence looking at anaphora and cataphora and Italian null and overt pronoun prefer- ences seems to be in conflict with Carminati’s PAH, indicating that null and overt